Modeling the commercial electricity demand in Santa Catarina, using the Box-Jenkins methodology

The present study was carried out to make forecasts of the monthly commercial electricity demand in the state of Santa Catarina. Historical data from January 2004 to December 2019 were used, with data from the last year being considered in the model validation process. After the exploratory analysis...

Descripción completa

Detalles Bibliográficos
Autores: Mattos, Viviane Leite Dias de, Nakamura, Luiz Ricardo, Konrath, Andréa Cristina, Bornia, Antônio Cezar
Tipo de recurso: artículo
Estado:Versión publicada
Fecha de publicación:2021
País:Brasil
Institución:Universidade Federal de Lavras (UFLA)
Repositorio:Repositório Institucional da UFLA
Idioma:inglés
OAI Identifier:oai:repositorio.ufla.br:1/50664
Acceso en línea:https://repositorio.ufla.br/handle/1/50664
Access Level:acceso abierto
Palabra clave:Electricity demand
Time series
Box-Jenkins Models
Forecasting
Demanda de eletricidade
Séries temporais
Modelos Box-Jenkins
Descripción
Sumario:The present study was carried out to make forecasts of the monthly commercial electricity demand in the state of Santa Catarina. Historical data from January 2004 to December 2019 were used, with data from the last year being considered in the model validation process. After the exploratory analysis based on descriptive measures, graphs and hypothesis tests, several other techniques were employed in the various stages of the Box-Jenkins methodology: identification, estimation, diagnosis and forecast. The SARIMA (1,1,1) (1,1,1)12 was selected as the one with the best performance according to some goodness-of-fit measures. In this model process, we identify some issues in the stationarity tests and in the use of both autocorrelation and partial autocorrelation functions used in the model identification. Nonetheless, other techniques, such as maximum likelihood estimation process, Ljung-Box, Jarque-Bera and ARCH for diagnostics and RMSE, MAPE and MAE as goodness-of-fit measures performed reasonable well, as expected. Only a few parameter values (zero up to three) of the Box-Jenkins models were considered in the model estimation stage. Practical implications: The fitted model can be used to provide electricity demand forecasting in the state of Santa Catarina, that may assist the planning of the electricity sector. Further, it may be used as subsidies on the development and improvement of public policies given that there is a great. In addition to fitting a proper model to represent the monthly commercial electricity demand in the State of Santa Catarina, we have identified some drawbacks in the applied methodology. Further studies may be performed to provide a better methodology and/or approach in order to obtain better and more accurateforecasts.